Static Tracker

1. A method comprising: scanning a response to a first HTTP request for a web page in order to identify a web object for prefetching; prefetching, from a content server, the web object prior to a browser requesting the web object; observing a plurality of child HTTP requests that are associated with the response to the first HTTP request for the web page; determining a measure of success of the prefetching, wherein the measure of success is based on whether the web object was prefetched and whether the web object was later requested by the browser; and modifying, based on the measure of success, a success rate for the web object as associated with the web page; wherein determining the measure of success comprises, after a duration of time during which an indication is not received that the web object was later requested, generating an inference that the web object was not requested by the browser, and wherein modifying the success rate is based on the inference.

2. The method of claim 1 further comprising deciding, based on the success rate for the web object as associated with the web page, whether to prefetch the web object based on the success rate.

3. The method of claim 1 further comprising: scanning the response to the first HTTP request for the web page in order to identify a second web object; determining, based on a second success rate for the second web object, not to prefetch the second web object; determining a second measure of success, wherein the second measure of success is based on whether the second web object was later requested by the browser; and modifying, based on the measure of success, the second success rate.

4. The method of claim 2 wherein determining the measure of success comprises recording, at the client side of the communications system, that the web object is subsequently requested by a user machine communicatively coupled with the client side of the communications system; and communicating a first message from the client side of the communications system to a server side of the communications system indicating that the web object is subsequently requested by the user machine.

5. The method of claim 4 wherein modifying the success rate is performed at the server side of the communications system according to the first message received from the client side of the communications system.

6. The method of claim 5 further comprising using a prefetch model, wherein the prefetch model comprises a plurality of bins, each corresponding to a set of web object characteristics; wherein the success rate is associated with a first bin of the plurality of bins; and wherein prefetching the web object according comprises: assigning the web object to the first bin of the plurality of bins by substantially matching characteristics of the web object to the set of object characteristics which correspond to the first bin; and determining whether to prefetch the object based on the success rate associated with the first bin of the plurality of bins.

7. The method of claim 6 , wherein updating the prefetch model at the server side of the communications system according to the message received from the client side of the communications system comprises: adjusting the success rate associated with the first bin of the plurality of bins according to the message.

8. The method of claim 6 , wherein the prefetch model represents URL scanner factors, user-specific URL tracker factors, and global URL tracker factors.

9. The method of claim 8 wherein the URL scanner factors comprise: a container value; and a tag value.

10. The method of claim 8 , wherein the global URL tracker factors comprise: a fails since success factor; and a number of successes factor.

11. The method of claim 6 wherein the prefetch model represents URL tracker factors such that the set of web object characteristics comprises a set of tracker factors; wherein the set of tracker factors comprise at least two of: a history of prefetch successes; a history of prefetch mistakes; a history of URL scanner misses; a page load age; and a scanner context associated with the web object for each page load.

12. The method of claim 6 , wherein the prefetch model is determined at least in part according to ages of prefetch objects represented by the prefetch model.

13. The method of claim 4 , wherein the first message indicates at least one of a hit, a miss, or a mistake.

14. A system comprising: a content server disposed at a server side of a communications system a client computer disposed at a client side of the communications system; a proxy server disposed at a server side of the communications system, and connected by communication links to the content server and the client computer, wherein the proxy server is configured to: scan a response from the content server to a first HTTP request for a web page from the client computer in order to identify a web object for prefetching; prefetch the web object prior to the client computer requesting the web object; observe a plurality of child HTTP requests from the client computer that are associated with the response to the first HTTP request for the web page from the content server; determine a measure of success of the prefetching, wherein the measure of success is based on whether the web object was prefetched and whether the web object was later requested by the browser; modify, based on the measure of success, a success rate for the web object as associated with the web page; wherein the processor is further configured to, after a duration of time during which an indication is not received that the web object was later requested, generate an inference that the web object was not later requested by the client side of the communications system, and wherein the processor is further configured to modify the success rate based on the inference.

15. The system of claim 14 wherein the proxy server determines the measure of success following receipt of a first message from the client computer indicating that the web object is requested by the user machine.

16. The system of claim 14 wherein the proxy server is further configured to update a prefetch model according to a first message received from the client computer.

17. The system of claim 14 , further comprising: a proxy client, in operative communication with the proxy server, and wherein the proxy client is configured to: record whether the prefetch object is subsequently requested by a user machine communicatively coupled with the proxy client; and communicate a first message to the proxy server indicating whether the prefetch object is subsequently requested by the user machine.

18. A non-transitory machine-readable medium for selective prefetch scanning in a communications system, the machine-readable medium having non-transitory instructions stored thereon which, when executed by a machine, cause the machine to perform steps, the program code comprising: program code for scanning a response to a first HTTP request for a web page in order to identify a web object for prefetching; program code for prefetching, from a content server, the web object prior to a browser requesting the web object; program code for observing a plurality of child HTTP requests that are associated with the response to the first HTTP request for the web page; program code for determining a measure of success of the prefetching, wherein the measure of success is based on whether the web object was prefetched and whether the web object was later requested by the browser; program code for modifying, based on the measure of success, a success rate for the web object as associated with the web page; wherein the program code for determining the measure of success comprises program code for, after a duration of time during which an indication is not received that the web object was later requested, generating an inference that the web object was not later requested by the client side of the communications system, and wherein the program code for modifying the success rate comprises program code for modifying the success rate based on the inference.

19. The non-transitory machine-readable medium of claim 18 , wherein the program code further comprises program code for deciding, based on the success rate for the web object as associated with the web page, whether to prefetch the web object based on the success rate.

20. The non-transitory machine-readable medium of claim 18 , wherein the program code further comprises: program code for scanning the response to the first HTTP request for the web page in order to identify a second web object; program code for determining, based on a second success rate, not to prefetch the second web object; program code for determining a second measure of success, wherein the second measure of success is based on whether the second web object was later requested by the browser; and program code for modifying, based on the measure of success, the second success rate.